Day 2: Quick Start
"Don't be intimidated by the word 'deploy.' If you can make a bowl of instant noodles, you can build an AI assistant. About the same amount of time—10 minutes."
Chapter Overview
Today you'll complete:
- Choosing your assistant's runtime environment (MyClaw Cloud or self-hosted)
- Getting your AI model API Key
- Connecting Telegram as your chat channel
- Sending your first message to your AI assistant
Today's Goal
By the end of today, you'll receive a message on your phone from your AI assistant.
Not someone else's assistant, not some platform's bot—your own, belonging only to you.
Ready? Let's begin.
Choose Your Runtime Environment
First, you need a place for your assistant to "live."
MyClaw Cloud: The fastest way to get started. Pick a plan (Lite, Pro, or Max), and your OpenClaw instance is provisioned automatically on dedicated cloud infrastructure. No server management, no SSH, no Linux commands. Your instance is online 24/7, auto-updated, and backed up. Just sign up, subscribe, and your assistant is ready in minutes.
Go to your MyClaw Dashboard to create your instance.
Self-hosted: Rent a cloud server (any VPS provider) or use a spare computer at home. You get full root access and complete control.
Recommended specs:
| Spec | Recommended |
|---|---|
| OS | Ubuntu 22.04 LTS |
| CPU | 2 cores |
| RAM | 4GB |
| Disk | 40GB SSD |
| Price | Starting from ~$5/month on most VPS providers |
Other options:
- Mac Mini / Old Laptop: Zero extra cost, data stays completely at home. Needs to stay powered on.
- Your Current Computer: Zero barrier, start immediately. Goes offline when you shut down—suitable for trying things out.
Preparation
Before setting up, prepare these two things—you'll need them regardless of which path you choose:
1. AI Model Access (choose one)
- If you have a Claude subscription (Pro/Max/Team), you can OAuth login during setup—no API Key needed
- Otherwise, go to console.anthropic.com to create an API Key (pay as you go)
Supported Models: Besides Anthropic Claude (recommended), OpenClaw also supports OpenAI GPT, Google Gemini, xAI Grok, and many others via OpenRouter. But Claude is the first choice for best results.
2. Telegram Bot Token
Open Telegram, search for @BotFather, send /newbot:
You: /newbot
BotFather: Alright, a new bot. How are we going to call it?
You: My AI Assistant
BotFather: Good. Now let's choose a username...
You: my_ai_assistant_bot
BotFather: Done! ... Use this token to access the HTTP API:
7234567890:AAHxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Copy this token and save it for later.
Why Telegram? Three reasons: (1) creating a Bot is free; (2) the API has the most features (supports buttons, files, voice...); (3) you can use it on any device.
Get Your Telegram User ID
In Telegram, search for @userinfobot—it will tell you your numeric ID. Note it down. This is used to restrict who can talk to your assistant, preventing others from using up your API quota.
Set Up Your Assistant
MyClaw Cloud: After subscribing, go to your Dashboard. Your OpenClaw instance is already provisioned. From the dashboard you can:
- Configure your AI model: Enter your API Key or connect via OAuth in the settings panel
- Connect Telegram: Paste your Bot Token and Telegram user ID in the channel configuration
- Start your instance: Click the start button — your assistant goes online immediately
The dashboard provides a visual interface for everything — no terminal commands needed. You can monitor your instance status, view logs, and manage configuration all from the web UI.
Self-hosted: Open your terminal (SSH into your server if using a cloud VPS) and run:
curl -fsSL https://openclaw.ai/install.sh | bash
That's it. One line. The install script handles all dependencies (Node.js, Git...), then launches the interactive setup wizard.
The wizard will ask you:
- Choose Mode: QuickStart (recommended) or Advanced
- Choose AI Model: Claude recommended. Connect via setup-token (if you have a subscription) or API Key
- Configure Chat Channel: Choose Telegram, paste the Bot Token
- Set Admin: Enter your Telegram user ID
- Install Background Daemon: Choose Yes — this runs your assistant in the background, starts on boot
If you're on Mac, install Homebrew first if you don't have it:
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
If you're on Windows, install WSL2 first (open PowerShell as Administrator, run wsl --install), then work inside WSL.
Your Assistant Is Online
MyClaw Cloud: Check your Dashboard — when the instance status shows "Running," your assistant is live. You can view real-time logs and health status directly from the dashboard.
Self-hosted: After the wizard completes, verify your assistant is running:
openclaw gateway status
If you see the Gateway is running, everything is ready.
Send Your First Message
Open Telegram, find the Bot you just created, send a message:
Hello! Who are you?
Wait a few seconds—you'll receive a reply.
This moment might not have fireworks or celebrations. But you just did something meaningful: you now own an AI assistant running on your own infrastructure. It's not a ChatGPT wrapper, not some platform's limited trial—it's completely, entirely yours.
Try chatting a bit more:
- "Write me a poem about cats"
- "What's the square root of 1024?"
- "Write a quicksort in Python"
- "What should I do today?" (It'll say something generic—but that changes on Day 3)
Right now, it's just an assistant that "can chat." But don't worry, over the next few days, we'll give it superpowers.
Daily Management
MyClaw Cloud: All management happens through your MyClaw Dashboard:
- Status monitoring: Real-time instance health and uptime
- Logs: View assistant activity and conversations
- Configuration: Change model, channels, and settings through the UI
- Restart: One-click restart from the dashboard if needed
Self-hosted: You'll use these commands frequently:
openclaw status # Check overall status
openclaw gateway status # Check Gateway running status
openclaw health # Health check
openclaw configure # Reconfigure (change model, channels, etc.)
openclaw daemon restart # Restart background service
openclaw daemon logs # View runtime logs
Troubleshooting Common Issues
Telegram Bot Not Responding
- Confirm Bot Token is correct
- Confirm your user ID is in the admin list
- Check logs (Dashboard for Cloud,
openclaw daemon logsfor self-hosted) - Confirm API Key is valid and has balance
- First DM might need pairing approval — check your dashboard or run
openclaw pairing list telegram
Want to Reconfigure?
MyClaw Cloud: Head to your Dashboard settings to change any configuration.
Self-hosted: Rerun the wizard anytime:
openclaw onboard
Or change a specific part:
openclaw configure
Key Takeaways
- Two paths to get started: MyClaw Cloud (instant, managed) or self-hosted (full control)
- Telegram Bot: Free to create, feature-rich API, works on any device
- Security first: Set admin ID so only you can chat with the assistant
- Always online: Whether cloud-managed or daemon-based, your assistant runs 24/7
- Next step: Give the assistant a soul — transform it from "generic AI" to "your AI"
Today's Achievement
Give yourself a pat on the back—today you completed:
- Chose your runtime environment
- Set up OpenClaw (via MyClaw Cloud or self-hosted installation)
- Created a Telegram Bot and connected successfully
- Successfully chatted with your AI assistant
You now have an AI assistant online 24 hours a day. Though it's still quite "generic" right now—like a new employee who just started, very capable but doesn't know you yet.
Tomorrow, we give it a soul.
Preview: Day 3 — Configure Your Assistant's Soul
Three files transform your assistant from "generic AI" to "your AI." SOUL.md defines personality, USER.md describes who you are, AGENTS.md sets its work habits. This is the most fun day of the entire 7 days—you'll create a one-of-a-kind AI character.
Ready to get started?
Get your own OpenClaw instance running on MyClaw Cloud in minutes.